Re: Adding new NSE discovered targets to Nmap

[David Fifield <david () bamsoftware com>]

On Fri, Aug 13, 2010 at 05:22:15PM +0100, Djalal Harouni wrote:
> On 2010-08-12 15:35:53 -0500, Ron wrote:
> > Attached a script that does a DHCP request then scans the DHCP address range. 
> >
> > I ran into a couple issues, though. 
> >
> > First, if I do socket:connect("255.255.255.255", 67, "udp") - it doesn't
> > work. It appears to, but packets don't get sent. I've run into issues with
> > broadcasting like that before, and I don't understand exactly what's going
> > on, but does anybody have any ideas? For now, in this script, I hardcoded
> > 192.168.1.1 as the router's address -- definitely not optimal. 
> >
> > Second, I have to use pcap_receive() to get the response from the DHCP
> > server (because DHCP is crazy). pcap_receive() requires an interface and a
> > MAC address. Since this is a prescan rule, there's no interface so I
> > hardcode "eth0". Is that going to break on non-Linux or is that standard
> > for Nmap? Is there some way I can enumerate ethernet interfaces so I can
> > send out the DHCP request on all of them? If not, would it be hard to add?
>
> We should add support for iflist and getinterfaces() functions to
> NSE to be able to enumerate and use net interfaces and perhaps routes ?
> First I've planned to pass them in the net table to prerule/postrule
> scripts, but the prefered solution was to not use a net table and only
> add some NSE API.

Yes, such an interface is a good idea. Do you want to make a design and
API for it? Or you, Ron? What do you need in order to make dhcp-discover
work? Hardcoding eth0 is definitely going to break in many situations.

I think what we need is a table that represents the same information as
struct interface_info (libnetutil/netutil.h). Like nmap.getinterfaces()
would return a table of such tables.

Henri showed that we can get these kinds of changes in fairly quickly.
I'd love to see a patch and then get in these script changes (once we
solve the braodcasting issue too).

David Fifield
_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev
Archived at http://seclists.org/nmap-dev/